Pixly: Live Photo AI Animator 4+

Asking for more money after you purchase the subscription

So I tried 5 pictures, the idea is nice but it’s unrealistic and scary. I used photo of my deceased grandfather and it turned quite scary, it also showed another person behind it. I tried different variations, but after the fifth it asked me to by credits 10 credits a for a photo (£5 for 50 credits). It doesn’t seem to me fair to have to pay for it once you are subscribed??? Also none of the pictures created are usable, therefore what am I paying for? To “try” and see if I get a decent one? I will ask Apple for refund. For me is not worth end

Not a scam, but does not work

I thought I would leave a review, there are lots of apps out there that are scams, I think this app is genuinely trying to be a good app, but it just does not work. My first photo I put in I was blown away, a black and white photo of my dad and I, brought to life! But then as I started to look closer at it, when my dad head turned it was not his face any more, I think the AI is getting images from the web or something and overlaying them and air brushing it in? Not sure. So I tried another couple of pics and they looked awful, the moment the person moved it was no longer their face, in fact quite disturbing. I was a bit annoyed being forced down the £6.99 route where I had just done 3 pics only to realise the app does not work, so I contacted apple for a refund, which I got no problem. So be careful with this one, the app really needs some work.

Developer Response ,

Hello! Sorry you had such a negative experience. Most of our clients are happy with the result. We will track your attempts and draw a conclusion about what went wrong.

Very promising

This App is fascinating and has lots of potential but it hasn’t been perfected yet, which is understandable because AI is still a young field. When it works, it works so well it’s mind blowing.

The flaws are as follows.

Lack of temporal coherency. Faces sometimes change.

Inability to correctly detect the depth of some objects. Understandable because 2D images don’t have depth information.

Disproportional limb sizes at times.

Inability to ascertain when a person is in mid flight motion. I gave it a picture of Michael Jordan in the middle of a dunk but he was transformed as if he was still.

Inability to understand geometric connectivity. It did not understand that a bra strap at the mid section of a woman’s arm was connected to the rest of the bra.

Unnatural body language of people sometimes when I gave it a picture of a large group.

It would also fail to animate background people when there were many foreground people to animate.

I look forward to seeing this technology improve over the years.

Developer Response ,

It's great to hear that you enjoyed it! We do continuously enhance the algorithm to make interacting with the bots more engaging and fulfill all your requests. If you have any suggestions or additional comments, feel free to leave them at: support@comapps.co